In case anyone was curious what it's based on. (No judgement either way from me, IMO depends how well they did the post-training, which I can't tell).
7,571 karma · joined April 20, 2018
In case anyone was curious what it's based on. (No judgement either way from me, IMO depends how well they did the post-training, which I can't tell).
The ~issue is though that they're using a pretty low quant, 2 bit. 4 bit are _pretty_ close to full ability, maybe degraded a bit. 2 bit is _noticably_ worse. It's about as small of a quant as you can get while still pretending it works, for this model.
This won't replace claude for you, unless you baby it a lot and are okay with it being kind or derpy.
2^n isn't a "fixed size" here.
Because you still get money?
Nothing about getting copied means you don't still get money. It probably just means your legal team has some people to annoy into going away if you're a big company.
That's just incorrect. Are you assuming the table starts pre-filled, and has O(1) lookup? Nobody is using a model like that for this problem.
2^n space complexity doesn't automatically mean you get to fill that space with 2^n chosen computed values for free.
There's models with "oracle"s for various sub-problems and such (that's essentially what you're describing as the ability you're baking in to your model of computation, it can look up the answer to any multiplication for free), but that's not an interesting model for this problem.
If you do that you're just asking "hey, how long does it take to multiply two numbers if multiplying two numbers is free". Boring question, and not the one anyone else is asking.
Well that's certainly not the claim. Think about it, that would mean the whole thing is trivially O(1), so why would they be bothering with the rest?
You've missed part in the analysis, or I believe added an assumption they're not making (the constant-time-multiply one, specifically).
This all _is_ almost certainly impractical, but it's for the usual reason of it only being faster for numbers nobody could ever care about. It's theoretically _quite_ interesting though, because hey...what _is_ the fastest you can actually multiply numbers if it's not O(n lg n) as many ~assumed.
Basically nothing else is like that. That does show the top tier of what companies can do if they try hard enough, but I don't think AI will get you there. I expect you still need someone to care _deeply_ to get everything just right to make it as difficult as possible, otherwise there's just so much room for one small mistake.
I'm not. When you're doing inference you typically reduce power limits anyway if you're optimizing for efficiency. Even if you don't, it doesn't spike to the max of the card.
My 3090 I power it down to 175W or so for single-digit reductions in performance for like, 60% of the wattage, something like that.
No. Certainly not now, maybe not ever. That is the least consumer-focused product page I've seen today.
I would bet it's mostly because it makes the humans uncomfortable.
They must still have human moderators for certain situations or those looking at the data for whatever reason. I can imagine it could be traumatizing to see what amounts to sustained verbal abuse without end.
I _think_ I expect it to be quite high. My understanding is it's "weak" because it falls off quickly, past a certain distance, due to the force carrier having mass.
Half of people seem to swear by them, and half hate them. They did seem great until I kept having to reconnect them every couple days, so I get why the lucky ones are happy.
Yeah, we're used to "convex" polygons. That's the phrasing to remove that part that feels wrong.
I have a couple of the Tuya ones and I'm pleased with them. The humidity accuracy is questionable, but it's fine.
I also have an airgradient (not zigbee, uses wifi) for high accuracy the one place I care about that. Works great in Home Assistant with esphome.
At least allow hardware+email overrides/allow-list. If someone has a banned cpu and they promise they're not the offender, why not allow them back on _with_ a fixed email address?
If they keep doing it, not like you won't notice by the time they're like "no really, the 4 other guys before me cheated, I bought this cpu from them".
> The joy comes from the journey and the experience.
> Look at competitive chess and Go and how they have fundamentally been transformed.
Go is better since AlphaGo. Tools are better, it's easier to learn from your games, we're better at it. The AI makes sick fucking moves and we get to see.
The journey is still there, the experience is still there.
Chess I doubt is worse off either, but I don't know chess that well.
You still need to get the backtracking correct and there's some minor tricks to work out, but that'll get you up to n=8 working quite well.